This thesis describes the design and implementation of a debugger for parallel programs executing on a system of loosely coupled processors. A primary goal has been to create a debugging environment that structurally matches the design of the distributed program. This means that the debugger supports hierarchical module structure, and communication flow through explicitly declared ports. The main advantages of our work over existing work in this area are: overview of the inter-process communication structure, a minimal amount of irrelevant information presented in the inspection tools, and support for working at different levels of detail. The debugging system consists of a trace collecting runtime component linked into the debugged program, and two window based tools for inspecting the traces. The debugger also includes animation of traced events.